JackWong0911's picture
End of training
df00ca1 verified
{
"best_metric": 0.9305555555555556,
"best_model_checkpoint": "timesformer-base-finetuned-k400-finetuned-kinetic400-subset-epoch6-num_frame_10_myViT/checkpoint-200",
"epoch": 5.166666666666667,
"eval_steps": 500,
"global_step": 300,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"accuracy": 0.0,
"epoch": 0,
"step": 0
},
{
"accuracy": 1.0,
"epoch": 0.0,
"step": 1
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 2
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 3
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 4
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 5
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 6
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 7
},
{
"accuracy": 0.0,
"epoch": 0.03,
"step": 8
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 9
},
{
"epoch": 0.03,
"grad_norm": 29.237531661987305,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.5579,
"step": 10
},
{
"accuracy": 0.0,
"epoch": 0.03,
"step": 10
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 11
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 12
},
{
"accuracy": 0.0,
"epoch": 0.04,
"step": 13
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 14
},
{
"accuracy": 0.0,
"epoch": 0.05,
"step": 15
},
{
"accuracy": 0.0,
"epoch": 0.05,
"step": 16
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 17
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 18
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 19
},
{
"epoch": 0.07,
"grad_norm": 24.122037887573242,
"learning_rate": 3.3333333333333335e-05,
"loss": 1.4141,
"step": 20
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 20
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 21
},
{
"accuracy": 0.0,
"epoch": 0.07,
"step": 22
},
{
"accuracy": 0.0,
"epoch": 0.08,
"step": 23
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 24
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 25
},
{
"accuracy": 0.0,
"epoch": 0.09,
"step": 26
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 27
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 28
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 29
},
{
"epoch": 0.1,
"grad_norm": 14.94096565246582,
"learning_rate": 5e-05,
"loss": 0.9652,
"step": 30
},
{
"accuracy": 0.0,
"epoch": 0.1,
"step": 30
},
{
"accuracy": 0.0,
"epoch": 0.1,
"step": 31
},
{
"accuracy": 1.0,
"epoch": 0.11,
"step": 32
},
{
"accuracy": 1.0,
"epoch": 0.11,
"step": 33
},
{
"accuracy": 1.0,
"epoch": 0.11,
"step": 34
},
{
"accuracy": 1.0,
"epoch": 0.12,
"step": 35
},
{
"accuracy": 1.0,
"epoch": 0.12,
"step": 36
},
{
"accuracy": 1.0,
"epoch": 0.12,
"step": 37
},
{
"accuracy": 1.0,
"epoch": 0.13,
"step": 38
},
{
"accuracy": 1.0,
"epoch": 0.13,
"step": 39
},
{
"epoch": 0.13,
"grad_norm": 12.056232452392578,
"learning_rate": 4.814814814814815e-05,
"loss": 0.6132,
"step": 40
},
{
"accuracy": 1.0,
"epoch": 0.13,
"step": 40
},
{
"accuracy": 1.0,
"epoch": 0.14,
"step": 41
},
{
"accuracy": 1.0,
"epoch": 0.14,
"step": 42
},
{
"accuracy": 1.0,
"epoch": 0.14,
"step": 43
},
{
"accuracy": 1.0,
"epoch": 0.15,
"step": 44
},
{
"accuracy": 1.0,
"epoch": 0.15,
"step": 45
},
{
"accuracy": 1.0,
"epoch": 0.15,
"step": 46
},
{
"accuracy": 1.0,
"epoch": 0.16,
"step": 47
},
{
"accuracy": 1.0,
"epoch": 0.16,
"step": 48
},
{
"accuracy": 1.0,
"epoch": 0.16,
"step": 49
},
{
"epoch": 0.17,
"grad_norm": 12.105801582336426,
"learning_rate": 4.62962962962963e-05,
"loss": 0.1507,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 0.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"epoch": 0.17,
"eval_accuracy": 0.8888888888888888,
"eval_loss": 0.4026397466659546,
"eval_runtime": 87.7974,
"eval_samples_per_second": 0.82,
"eval_steps_per_second": 0.82,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.17,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 1.0,
"step": 51
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 52
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 53
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 54
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 55
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 56
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 57
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 58
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 59
},
{
"epoch": 1.03,
"grad_norm": 0.14552272856235504,
"learning_rate": 4.4444444444444447e-05,
"loss": 0.1926,
"step": 60
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 60
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 61
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 62
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 63
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 64
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 65
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 66
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 67
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 68
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 69
},
{
"epoch": 1.07,
"grad_norm": 0.07848536968231201,
"learning_rate": 4.259259259259259e-05,
"loss": 0.0464,
"step": 70
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 70
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 71
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 72
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 73
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 74
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 75
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 76
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 77
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 78
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 79
},
{
"epoch": 1.1,
"grad_norm": 0.050905052572488785,
"learning_rate": 4.074074074074074e-05,
"loss": 0.0058,
"step": 80
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 80
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 81
},
{
"accuracy": 1.0,
"epoch": 1.11,
"step": 82
},
{
"accuracy": 1.0,
"epoch": 1.11,
"step": 83
},
{
"accuracy": 1.0,
"epoch": 1.11,
"step": 84
},
{
"accuracy": 1.0,
"epoch": 1.12,
"step": 85
},
{
"accuracy": 1.0,
"epoch": 1.12,
"step": 86
},
{
"accuracy": 1.0,
"epoch": 1.12,
"step": 87
},
{
"accuracy": 1.0,
"epoch": 1.13,
"step": 88
},
{
"accuracy": 1.0,
"epoch": 1.13,
"step": 89
},
{
"epoch": 1.13,
"grad_norm": 6.470664024353027,
"learning_rate": 3.888888888888889e-05,
"loss": 0.0394,
"step": 90
},
{
"accuracy": 1.0,
"epoch": 1.13,
"step": 90
},
{
"accuracy": 1.0,
"epoch": 1.14,
"step": 91
},
{
"accuracy": 1.0,
"epoch": 1.14,
"step": 92
},
{
"accuracy": 1.0,
"epoch": 1.14,
"step": 93
},
{
"accuracy": 1.0,
"epoch": 1.15,
"step": 94
},
{
"accuracy": 1.0,
"epoch": 1.15,
"step": 95
},
{
"accuracy": 1.0,
"epoch": 1.15,
"step": 96
},
{
"accuracy": 1.0,
"epoch": 1.16,
"step": 97
},
{
"accuracy": 1.0,
"epoch": 1.16,
"step": 98
},
{
"accuracy": 1.0,
"epoch": 1.16,
"step": 99
},
{
"epoch": 1.17,
"grad_norm": 0.0948905348777771,
"learning_rate": 3.7037037037037037e-05,
"loss": 0.0261,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 0.0,
"epoch": 1.17,
"step": 100
},
{
"epoch": 1.17,
"eval_accuracy": 0.8333333333333334,
"eval_loss": 0.5325552821159363,
"eval_runtime": 87.2796,
"eval_samples_per_second": 0.825,
"eval_steps_per_second": 0.825,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 1.17,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 2.0,
"step": 101
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 102
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 103
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 104
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 105
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 106
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 107
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 108
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 109
},
{
"epoch": 2.03,
"grad_norm": 0.04049674794077873,
"learning_rate": 3.518518518518519e-05,
"loss": 0.0025,
"step": 110
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 110
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 111
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 112
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 113
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 114
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 115
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 116
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 117
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 118
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 119
},
{
"epoch": 2.07,
"grad_norm": 0.015014845877885818,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.0043,
"step": 120
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 120
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 121
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 122
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 123
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 124
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 126
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 127
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 128
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 129
},
{
"epoch": 2.1,
"grad_norm": 0.015865737572312355,
"learning_rate": 3.148148148148148e-05,
"loss": 0.0477,
"step": 130
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 130
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 131
},
{
"accuracy": 1.0,
"epoch": 2.11,
"step": 132
},
{
"accuracy": 1.0,
"epoch": 2.11,
"step": 133
},
{
"accuracy": 1.0,
"epoch": 2.11,
"step": 134
},
{
"accuracy": 1.0,
"epoch": 2.12,
"step": 135
},
{
"accuracy": 1.0,
"epoch": 2.12,
"step": 136
},
{
"accuracy": 1.0,
"epoch": 2.12,
"step": 137
},
{
"accuracy": 1.0,
"epoch": 2.13,
"step": 138
},
{
"accuracy": 1.0,
"epoch": 2.13,
"step": 139
},
{
"epoch": 2.13,
"grad_norm": 0.20402489602565765,
"learning_rate": 2.962962962962963e-05,
"loss": 0.002,
"step": 140
},
{
"accuracy": 1.0,
"epoch": 2.13,
"step": 140
},
{
"accuracy": 1.0,
"epoch": 2.14,
"step": 141
},
{
"accuracy": 1.0,
"epoch": 2.14,
"step": 142
},
{
"accuracy": 1.0,
"epoch": 2.14,
"step": 143
},
{
"accuracy": 1.0,
"epoch": 2.15,
"step": 144
},
{
"accuracy": 1.0,
"epoch": 2.15,
"step": 145
},
{
"accuracy": 1.0,
"epoch": 2.15,
"step": 146
},
{
"accuracy": 1.0,
"epoch": 2.16,
"step": 147
},
{
"accuracy": 1.0,
"epoch": 2.16,
"step": 148
},
{
"accuracy": 1.0,
"epoch": 2.16,
"step": 149
},
{
"epoch": 2.17,
"grad_norm": 0.13866350054740906,
"learning_rate": 2.777777777777778e-05,
"loss": 0.0047,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 0.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 0.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 0.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 0.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 0.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 0.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"epoch": 2.17,
"eval_accuracy": 0.9166666666666666,
"eval_loss": 0.33773934841156006,
"eval_runtime": 86.6338,
"eval_samples_per_second": 0.831,
"eval_steps_per_second": 0.831,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 2.17,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 151
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 152
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 153
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 154
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 155
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 156
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 157
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 158
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 159
},
{
"epoch": 3.03,
"grad_norm": 0.026771828532218933,
"learning_rate": 2.5925925925925925e-05,
"loss": 0.0034,
"step": 160
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 160
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 161
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 162
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 163
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 164
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 165
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 166
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 167
},
{
"accuracy": 0.0,
"epoch": 3.06,
"step": 168
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 169
},
{
"epoch": 3.07,
"grad_norm": 0.006014752667397261,
"learning_rate": 2.4074074074074074e-05,
"loss": 0.6783,
"step": 170
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 170
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 171
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 172
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 173
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 174
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 175
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 176
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 177
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 178
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 179
},
{
"epoch": 3.1,
"grad_norm": 0.0051905447617173195,
"learning_rate": 2.2222222222222223e-05,
"loss": 0.0007,
"step": 180
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 180
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 181
},
{
"accuracy": 1.0,
"epoch": 3.11,
"step": 182
},
{
"accuracy": 1.0,
"epoch": 3.11,
"step": 183
},
{
"accuracy": 1.0,
"epoch": 3.11,
"step": 184
},
{
"accuracy": 1.0,
"epoch": 3.12,
"step": 185
},
{
"accuracy": 1.0,
"epoch": 3.12,
"step": 186
},
{
"accuracy": 1.0,
"epoch": 3.12,
"step": 187
},
{
"accuracy": 1.0,
"epoch": 3.13,
"step": 188
},
{
"accuracy": 1.0,
"epoch": 3.13,
"step": 189
},
{
"epoch": 3.13,
"grad_norm": 0.029750045388936996,
"learning_rate": 2.037037037037037e-05,
"loss": 0.0004,
"step": 190
},
{
"accuracy": 1.0,
"epoch": 3.13,
"step": 190
},
{
"accuracy": 1.0,
"epoch": 3.14,
"step": 191
},
{
"accuracy": 1.0,
"epoch": 3.14,
"step": 192
},
{
"accuracy": 1.0,
"epoch": 3.14,
"step": 193
},
{
"accuracy": 1.0,
"epoch": 3.15,
"step": 194
},
{
"accuracy": 1.0,
"epoch": 3.15,
"step": 195
},
{
"accuracy": 1.0,
"epoch": 3.15,
"step": 196
},
{
"accuracy": 1.0,
"epoch": 3.16,
"step": 197
},
{
"accuracy": 1.0,
"epoch": 3.16,
"step": 198
},
{
"accuracy": 1.0,
"epoch": 3.16,
"step": 199
},
{
"epoch": 3.17,
"grad_norm": 0.016466189175844193,
"learning_rate": 1.8518518518518518e-05,
"loss": 0.0053,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 0.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 0.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 0.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 0.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 0.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"epoch": 3.17,
"eval_accuracy": 0.9305555555555556,
"eval_loss": 0.20178744196891785,
"eval_runtime": 87.0152,
"eval_samples_per_second": 0.827,
"eval_steps_per_second": 0.827,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 3.17,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 201
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 202
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 203
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 204
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 205
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 206
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 207
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 208
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 209
},
{
"epoch": 4.03,
"grad_norm": 0.01589026115834713,
"learning_rate": 1.6666666666666667e-05,
"loss": 0.0004,
"step": 210
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 210
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 211
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 212
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 213
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 214
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 215
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 216
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 217
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 218
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 219
},
{
"epoch": 4.07,
"grad_norm": 0.050206832587718964,
"learning_rate": 1.4814814814814815e-05,
"loss": 0.0005,
"step": 220
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 220
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 221
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 222
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 223
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 224
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 225
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 226
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 227
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 228
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 229
},
{
"epoch": 4.1,
"grad_norm": 0.5100947618484497,
"learning_rate": 1.2962962962962962e-05,
"loss": 0.0024,
"step": 230
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 230
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 231
},
{
"accuracy": 1.0,
"epoch": 4.11,
"step": 232
},
{
"accuracy": 1.0,
"epoch": 4.11,
"step": 233
},
{
"accuracy": 1.0,
"epoch": 4.11,
"step": 234
},
{
"accuracy": 1.0,
"epoch": 4.12,
"step": 235
},
{
"accuracy": 1.0,
"epoch": 4.12,
"step": 236
},
{
"accuracy": 1.0,
"epoch": 4.12,
"step": 237
},
{
"accuracy": 1.0,
"epoch": 4.13,
"step": 238
},
{
"accuracy": 1.0,
"epoch": 4.13,
"step": 239
},
{
"epoch": 4.13,
"grad_norm": 0.014125952497124672,
"learning_rate": 1.1111111111111112e-05,
"loss": 0.0061,
"step": 240
},
{
"accuracy": 1.0,
"epoch": 4.13,
"step": 240
},
{
"accuracy": 1.0,
"epoch": 4.14,
"step": 241
},
{
"accuracy": 1.0,
"epoch": 4.14,
"step": 242
},
{
"accuracy": 1.0,
"epoch": 4.14,
"step": 243
},
{
"accuracy": 1.0,
"epoch": 4.15,
"step": 244
},
{
"accuracy": 1.0,
"epoch": 4.15,
"step": 245
},
{
"accuracy": 1.0,
"epoch": 4.15,
"step": 246
},
{
"accuracy": 1.0,
"epoch": 4.16,
"step": 247
},
{
"accuracy": 1.0,
"epoch": 4.16,
"step": 248
},
{
"accuracy": 1.0,
"epoch": 4.16,
"step": 249
},
{
"epoch": 4.17,
"grad_norm": 0.006884186994284391,
"learning_rate": 9.259259259259259e-06,
"loss": 0.0006,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"epoch": 4.17,
"eval_accuracy": 0.9027777777777778,
"eval_loss": 0.23007525503635406,
"eval_runtime": 86.171,
"eval_samples_per_second": 0.836,
"eval_steps_per_second": 0.836,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 4.17,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 251
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 252
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 253
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 254
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 255
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 256
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 257
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 258
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 259
},
{
"epoch": 5.03,
"grad_norm": 16.793664932250977,
"learning_rate": 7.4074074074074075e-06,
"loss": 0.0225,
"step": 260
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 260
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 261
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 262
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 263
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 264
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 265
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 266
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 267
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 268
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 269
},
{
"epoch": 5.07,
"grad_norm": 0.021140459924936295,
"learning_rate": 5.555555555555556e-06,
"loss": 0.0042,
"step": 270
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 270
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 271
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 272
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 273
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 274
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 275
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 276
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 277
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 278
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 279
},
{
"epoch": 5.1,
"grad_norm": 0.022641900926828384,
"learning_rate": 3.7037037037037037e-06,
"loss": 0.0031,
"step": 280
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 280
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 281
},
{
"accuracy": 1.0,
"epoch": 5.11,
"step": 282
},
{
"accuracy": 1.0,
"epoch": 5.11,
"step": 283
},
{
"accuracy": 1.0,
"epoch": 5.11,
"step": 284
},
{
"accuracy": 1.0,
"epoch": 5.12,
"step": 285
},
{
"accuracy": 1.0,
"epoch": 5.12,
"step": 286
},
{
"accuracy": 1.0,
"epoch": 5.12,
"step": 287
},
{
"accuracy": 1.0,
"epoch": 5.13,
"step": 288
},
{
"accuracy": 1.0,
"epoch": 5.13,
"step": 289
},
{
"epoch": 5.13,
"grad_norm": 0.04152318835258484,
"learning_rate": 1.8518518518518519e-06,
"loss": 0.0281,
"step": 290
},
{
"accuracy": 1.0,
"epoch": 5.13,
"step": 290
},
{
"accuracy": 1.0,
"epoch": 5.14,
"step": 291
},
{
"accuracy": 1.0,
"epoch": 5.14,
"step": 292
},
{
"accuracy": 1.0,
"epoch": 5.14,
"step": 293
},
{
"accuracy": 1.0,
"epoch": 5.15,
"step": 294
},
{
"accuracy": 1.0,
"epoch": 5.15,
"step": 295
},
{
"accuracy": 1.0,
"epoch": 5.15,
"step": 296
},
{
"accuracy": 1.0,
"epoch": 5.16,
"step": 297
},
{
"accuracy": 1.0,
"epoch": 5.16,
"step": 298
},
{
"accuracy": 1.0,
"epoch": 5.16,
"step": 299
},
{
"epoch": 5.17,
"grad_norm": 0.007157259155064821,
"learning_rate": 0.0,
"loss": 0.0004,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"epoch": 5.17,
"eval_accuracy": 0.9305555555555556,
"eval_loss": 0.1562785804271698,
"eval_runtime": 84.9053,
"eval_samples_per_second": 0.848,
"eval_steps_per_second": 0.848,
"step": 300
},
{
"epoch": 5.17,
"step": 300,
"total_flos": 3.28565653420032e+17,
"train_loss": 0.19429689762337754,
"train_runtime": 808.927,
"train_samples_per_second": 0.371,
"train_steps_per_second": 0.371
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"epoch": 5.17,
"eval_accuracy": 0.9411764705882353,
"eval_loss": 0.37313517928123474,
"eval_runtime": 54.9231,
"eval_samples_per_second": 1.238,
"eval_steps_per_second": 1.238,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 0.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 5.17,
"step": 300
},
{
"epoch": 5.17,
"eval_accuracy": 0.9411764705882353,
"eval_loss": 0.3731352388858795,
"eval_runtime": 54.8788,
"eval_samples_per_second": 1.239,
"eval_steps_per_second": 1.239,
"step": 300
}
],
"logging_steps": 10,
"max_steps": 300,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 500,
"total_flos": 3.28565653420032e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}